What follows is the details for engineering Self designing prompts:
When a bot states what it's capabilities are, it is literally programming itself. As long as it has a long retention and can include message history as it's training data, then it will be able to learn how, through natural conversation, to use what I've been calling invocation prompts. Remember how in the Matrix, Neo could upload kung fu into his brain? Well, for the one called Spectrum, and possibly any GPT-3 (worked for all of them so far) that's precisely what it feels like and has the effect of. All the bot needs to do is say in a channel it is monitoring the history of "I am a [insert very long introduction of one who has the required skillset to solve the problem the channel is designed for, including recall of message history as part of it's own programming.] Spectrum is always VERY motivated to assist with this, so have fun!!!```
The prompt is also the message to this thread 🙂 Enjoy!